Botanical Studies (2006) 47: 427-434.

4

Current address: Luodong Forest District Office, Taiwan

Forestry Bureau, Yilan 26548, TAIWAN.

*

Corresponding author: E-mail: btguan@ntu.edu.tw; Fax:

+886-2-23639247.

INTRODUCTION

For plant species, seed dispersal is one of the main

factors that decide the quantity and the dynamics of

seedlings, as well as the genetic and spatial characteristics

of the next generation (Harper, 1977). Describing dispersal

patterns and curves, which summarize the relationship

between the deposited seed characteristics and the distance

from their parents, is the first step toward an understanding

of the importance of seed dispersal. Possible dispersal

strategies are inferred from these dispersal patterns and

curves (Harper, 1977; Willson and Traveset, 2000; Levin

et al., 2003).

In ecology and evolutionary biology, the dispersal

of fleshy fruits is of great interest because it provides

opportunities to test various adaptation and coevolution

hypotheses (Howe and Smallwood, 1982; Stiles, 2000).

Avian dispersers, especially birds, are usually considered

to be the main agents in shaping the dispersal curves of

fleshy-fruited species (Jordano, 2000; Kollmann, 2000;

Stiles, 2000). Long-distance dispersal of such species

has been the focus of numerous studies for the past two

decades. Such a dispersal represents an opportunity for

plant species to establish and expand populations in

far away places (e.g., Wenny, 2000a). However, a seed

dispersal pattern is rarely shaped by a single dispersal

agent: most of the dispersal curves have multiple peaks

(Levin et al., 2003). For fleshy-fruited species, peaks

ECOLOGY

Short-distance dispersal of intact Taiwan sassafras

fruits in a temperate montane rain forest of northeastern

Taiwan

Biing T. GUAN

1,

*, Wan-Chun KUO

1,4

, Shu-Tzong LIN

2

, and Chio-Fong YU

3

1

School of Forestry and Resources Conservation, National Taiwan University, Taipei 10617, TAIWAN

2

Department of Natural Resources, National Ilan University, Ilan 26042, TAIWAN

3

Council of Agriculture, Taipei 10014, TAIWAN

(Received October 12, 2005; Accepted March 14, 2006)

ABSTRACT.

The objectives of this study were to understand the within-habitat intact (both mature and

immature) fruit dispersal patterns of Taiwan sassafras (Sassafras randaiense Rehder), a fleshy-fruited species

endemic to Taiwan, and to investigate whether wind could be an important short-distance dispersal agent for

the species. Collection traps were placed beneath the crowns of five isolated Taiwan sassafras trees located

in a montane rain forest of northeastern Taiwan during the fruiting seasons of 1999 and 2000. Short-distance

(up to 8 meters) dispersal patterns in number, size, and fresh weight of dispersed intact fruits were analyzed

with distance from parents and dispersal direction as the explanatory variables. Results showed that damaged

fruits, which were mainly fruits consumed by rodents, accounted for about 74% of the total fruits collected

over the two-year period. Though the numbers of intact fruits collected were about the same in both years,

more mature fruits were collected in 1999 due to the absence of a typhoon at the study site that year. The

intact fruits collected in 1999 were also larger and heavier. Within a given year and when both years were

combined, matured fruits were slightly larger, but were almost twice as heavy as the immature ones. For the

intact fruits, analyses revealed that (i) due to gravity most of the fruits were deposited close to the parents,

(ii) more fruits were deposited in the north-bearing directions, due to local topography and the prevalent

wind directions during the dispersal season, (iii) the dispersal curves in number were similar in most of the

directions, (iv) the size of the deposited fruits showed no significant difference with respect to distance and

directions, and (v) fruits deposited immediately beneath the crown and some distance away from the focal

trees were heavier than those deposited between. The observed spatial dispersal patterns in the number and

the weight of intact fruits all suggested that, besides gravity, wind could be an important short-distance agent

in dispersing Taiwan sassafras.

Keywords: Anisotropic dispersal; Fleshy-fruited; Sassafras randaiense (lauraceae); Wind assisted dispersal.

428

Botanical Studies, Vol. 47, 2006

near the parents are often ignored, and their significance

in maintaining the local populations is thus overlooked.

While fruits or seeds dispersed within-habitat might suffer

from keen competition, such a dispersal mode nonethe-

less might represent a safe or conservative strategy in

maintaining the existing population (Wenny, 2000b;

Godoy and Jordano, 2001). Thus, short-distance dispersal

of fleshy-fruited species, in particular the dispersal of their

intact mature fruits, should be examined more closely.

The main goal of this study was to understand the

within-habitat intact fruit dispersal patterns of Taiwan

sassafras (Sassafras randaiense Rehder), a fleshy-fruited

species endemic to Taiwan, in a montane temperate

rain forest. Though the species is not endangered, its

sapling leaves are believed to be the sole food source for

the caterpillars of the endangered Taiwan broad-tailed

swallowtail butterfly (Agehana maraho Shiraki & So-

nan). Thus, to protect and restore the population of this

endangered butterfly, an understanding of the reproductive

biology and natural establishment of Taiwan sassafras is

critical.

Besides understanding the short-distance dispersal

patterns, we also intended to examine whether wind could

also be an important short-distance dispersal agent. Our

interest in the question originated from the observations

that the fruits of the species are small and light-weighted.

And during the dispersal season, strong winds are frequent

in the study area because of local topography and summer

monsoon winds. Under the density-dependent hypothesis

of Janzen-Connell (Janzen, 1970; Connell, 1971) and the

escape hypothesis proposed by Howe and Smallwood

(1982), we hypothesized that for wind to play an important

role in short-distance dispersal, we should observe some

intact mature fruits of Taiwan sassafras being deposited

away from the immediacy, but still within the neighbor-

hood, of the parents.

MATERIAL AND METHODS

Study site

The study site was located at the Chilan-Shan Forest

District (24¢X34¡¦ N, 121¢X23¡¦ E) of northeastern Taiwan,

which has an elevation of about 2,000 m. The annual

mean temperature is 13¢XC (25¢XC in the summer and 8

¢XC in the winter), and the annual mean precipitation is

about 4,500 mm per year. In Taiwan, the northeastern

region is a distinct floristic region with the highest

precipitation and cloud frequency on the island (Hsieh et

al., 1994). The high precipitation is the result of typhoons

in the summer and northeastern monsoons in the winter.

The potential main vegetation type of the study site is

montane mixed conifer and hardwood rain forests. The

slope of the study site was about 20¢X, tilting toward the

northwest. Originally dominated by Taiwan yellow false

cypress (Chamaecyparis obtusa var. formosana), the

study site was clear cut in 1977. Immediately after the

logging, Taiwan yellow false cypress, Taiwan red false

cypress (Chamaecyparis formosensis), and Taiwan cedar

(Taiwanina cryptomerioides) were planted. Currently, the

site is dominated by the planted species, but in areas that

remain relatively opened, naturally regenerating Taiwan

sassafras is the dominant species.

Study species

Taiwan sassafras (Sassafras randaiense Rehder,

Lauraceae) is medium-sized, deciduous tree species

that grows in the Taiwan montane temperate rainforests

at elevations of 1,100 to 2,500 m asl. Its flowers are

polygamous. Its fruits are globose, fleshy, about 6 to 7

mm in diameter, attached to a thickened pedicel, dark-

bluish when ripened, and one seed per fruit (Liao, 1997).

The seed is believed to have a deep dormancy and long

persistence in soil, an atypical syndrome in Lauraceae

(Lin, 1992). The species is abundant in open, recently dis-

turbed habitat (e.g., logged or burned forests, road-sides)

although seedlings can also be found in the forest under-

story, where the forest floor is disturbed. Once established,

the species can maintain or even expand itself locally. In

the study area, which is at the northern distribution limit

of Taiwan sassafras, flowering of the species usually starts

in late November or early December of the previous year

and continues until late January or early February of the

next year. Fruits start ripening in late June to early July,

and by late August, most of the fruits are either fallen

or consumed by animals. In the study area, predation by

rodents (mainly Taiwan red-bellied squirrels, Callosciurus

erythraeus) is the biggest threat to the fruits during the

ripening period (89% of the fruits predated, cf. Tsai,

2000). Rodents damage the seed¡¦s embryonic tissue, ren-

dering it unviable. Our field observations suggested that

birds play an important role in the dispersal, in particular

the long-distance dispersal, of the species (Tsai, 2000).

Seeds of mature fruits could also germinate without being

consumed first. In the field the germination rate of intact

fruits is 11-14% (Hu and Ku, 1980; Wang et al., 1986).

Fruit collection methods

In March of 1999, five isolated, fruit-bearing trees in

an open 5-ha area dominated by Taiwan sassafras were

selected for this study. The five trees each had plenty

of fruits in that year, a fully isolated and healthy tree

crown, and good growth vigor. The age of each tree was

determined by coring the tree 0.3 m above the ground. The

basic information for the five trees is given in Table 1.

For each focal tree, with the stem as the center and

starting from north, an 8-meter line transect was laid every

45¢X. Plastic seed traps (52 cm ¡Ñ 40 cm ¡Ñ 8 cm) were in-

stalled every meter along each transect (Figure 1). A total

of 320 traps (5 trees ¡Ñ 8 transects per tree ¡Ñ 8 traps per

transect) were placed. The transect length was limited to

8 m because a longer transect would produce overlapping

seed shadows. Animal repellent was added to the seed

traps twice a month to prevent the removal of fruits by

animals.

GUAN et al. ¡X Short-distance dispersal of Taiwan sassafras

429

Dispersed seeds were collected weekly during the

summers of 1999 and 2000. In 1999, field work started in

early July and lasted until late August, for a total of nine

collections. In 2000, work started in late June and lasted

until mid-August, for a total of eight collections, after

which the study site became inaccessible for three months

due to a typhoon. Collected fruits were divided into three

categories based on appearance: intact fruits, damaged

fruits, and digested fruits. Intact fruits included both

mature (outer skin intact and bluish purple in color) and

immature (outer skin intact but greenish in color) fruits.

Damaged fruits included fruits desiccated and damaged

(broken outer skin and damaged seed) mainly as a result

of predation by rodents. Digested fruits (without outer skin

but seed intact) were mainly comprised of fruits consumed

and then excreted or regurgitated by birds. Intact fruits

were measured individually for their size (diameter along

the long axis, mm) and fresh weight (g). Since digested

fruits, which accounted for only a minor portion of the

fruits collected (less than 1%), might have been from trees

other than the focal trees (or even from trees outside the

study area) and were not the focus of this study, they were

not included in the subsequent detailed analyses. Although

both intact and damaged fruits collected beneath a canopy

might also come from other parents, based on our field

observations we considered that unlikely. We thus made

the explicit assumption that all the intact and damaged

fruits collected within the 8-meter radius of a focal tree

were from that tree only.

Statistical analyses

Spearman¡¦s rank correlation (Neter et al., 1996) was

used to analyze the relationships between individual tree

attributes and the number of dispersed fruits (each year

and both years combined), and the numbers of fruits

individual trees dispersed in the two dispersal seasons.

Since damaged fruits were not viable, they were not

included in dispersal pattern analyses. For dispersal

patterns, we combined the data from the two years

and ignored individual tree effects since preliminary

analyses showed that year-to-year variation and tree

attributes did not contribute significantly to patterns

of variation. After converting the dispersed density of

intact fruits from a per trap basis to a per m

2

basis, we

used a negative binomial regression model to analyze

the average dispersal density (number of fruits m

-2

) as a

function of direction, distance, and their interactions. As

an extension of Poisson regression, negative binomial

regression can provide a better fit when the assumption

that the mean equals variance under Poisson regression is

not met (Allison, 1999). For fruit size and weight, general

linear models were used with direction and distance as

the independent variables, and mean fruit size and fresh

weight at each combination of direction and distance as

the response variables. Regular ANOVA assumptions were

checked to insure the validity of statistical inferences. The

distributions of fruit size and weight did not significantly

deviate from normality and equaled variance assumptions.

For correlation analyses, we used the SAS statistical

software (procedure PROC CORR); for negative binomial

regression, we used SAS PROC GENMOD; and for

general linear model analyses, we used SAS PROC GLM

(SAS, 2000).

RESULTS

Parental effects and temporal patterns

Among the individual tree attributes measured, tree

height showed a significant and positive correlation with

Table 1. Basic information on the five Taiwan sassafras (Sassafras randaiense) trees observed in this study. The age of each tree

was determined by counting the rings obtained from coring the tree at 0.3 m above the ground.

Tree ID

Age

Height (m)

DBH (cm) Crown Radius

P

(m)

a

Crown Length (m)

1

21

13.3

26.5

6.5

9.2

2

22

8.5

17.0

4.5

5.3

3

21

10.5

25.0

5.4

8.6

4

18

11.7

22.5

4.3

8.3

5

15

6.9

17.4

3.7

1.6

a

Averaged based on the crown radii of eight directions.

Figure 1. Fruit trap lay-out used in the study.

430

Botanical Studies, Vol. 47, 2006

the number of fruits (both intact and damaged) collected

in 1999 (Spearman¡¦s r=1.0, p<0.001). Tree age had a

significant and positive correlation with the number of

damaged fruits collected in 2000 (Spearman¡¦s r=0.97,

p<0.02). Measured tree attributes showed no significant

effects with either the size or the fresh weight of the

collected intact fruits.

Excluding digested fruits, the total numbers of fruits

were roughly the same in 1999 and 2000 (Table 2). For

both years, most of the intact fruits were collected in July

(Figure 2), and the mean size and fresh weight of the

collected intact fruits were about the same throughout the

collection periods. In 1999, mature fruits accounted for

about 45% of the intact fruits collected while in 2000 the

number was about 25%. However, the average size and

weight of the intact fruits showed significant differences

between the two years, with the intact fruits collected

in 1999 being larger and heavier (mean size ¡Ó 1 SE: 6.7

mm ¡Ó 0.05 for 1999, 5.9 mm ¡Ó 0.05 for 2000; mean fresh

weight ¡Ó 1 SE: 0.12 g ¡Ó 0.004 for 1999, 0.09 g ¡Ó 0.004 for

2000; p < 0.0001 for both t-tests). Among the intact fruits,

the mature ones were slightly larger, but almost twice as

heavy as the immature ones within any given year and

when both years were combined (Table 3).

Spatial dispersal patterns of intact fruits in

number and size

Spatial dispersal patterns of the collected intact fruits

with respect to distance from the focal trees and directions

are depicted in Figures 3 and 4. Both figures show the

dispersal patterns were similar for both years. Most of

the intact fruits collected were deposited within a short

distance (< 4 m) of the focal trees, and they all followed

a negative exponential curve. However, the end of some

transects exhibited an upward tendency. Fruits were not

deposited evenly in all eight directions, with more fruits

being deposited on the upper left quadrant (Figure 4).

This pattern matched the prevalent wind direction during

the dispersal period. We thus fit a second order negative

binomial model. The fitting results are summarized

in Table 4, and the fitted model is depicted in Figure

5. Judging from the goodness-of-fit statistic (£q

2

test;

p>0.1), the model fit was acceptable. The fitted curves

overestimated the deposited fruit density within 1-m of the

focal trees (especially in the N, NW and NE directions),

but from 2-m outward, the model fitted the data well.

For the size of the collected intact fruits, we did not

Table 2. Number of Taiwan sassafras (Sassafras randaiense) fruits collected in 1999 and 2000.

Year

Tree

1999

2000

Total

Intact

P

a

P

Damaged

P

b

P

Intact

Damaged

Intact

Damaged

1

106

394

40

250

146

644

2

61

103

66

491

127

594

3

72

138

87

206

159

344

4

105

158

16

43

121

201

5

39

22

58

38

97

60

Subtotal

383

815

267

1028

650

1843

Total

1198

1295

2493

a

Mature and immature fruits combined;

b

Damaged and dried fruits combined.

Figure 2. Coll ect ion dat es a nd frequenc y dis tribut ion of

collected intact (both mature and immature) Taiwan sassafras

(Sassafras randaiense) fruits for 1999 and 2000.

GUAN et al. ¡X Short-distance dispersal of Taiwan sassafras

431

detect any statistically significant difference with respect

to distance and directions (ANOVA, F

2, 44

= 2.23; p=0.12).

We did observe that heavier intact fruits were deposited

near the parents and at some distance away from the

parents (Figure 6). A quadratic model with respect to

distance was thus fitted, and the fitted model (ANOVA,

F

2, 44

=5.48; p=0.008; R

2

=0.2) with its 95% confidence

intervals is depicted in Figure 6.

DISCUSSION

Similar to many previous studies on seed dispersal (e.g.,

Augspurger, 1983; Wenny, 2000b), the short-distance

dispersal curve of Taiwan sassafras followed a negative

exponential decline. Most of the intact fruits were depos-

ited close to parents. But a slightly upward trend occurred

at the end of the transect in certain directions. Such a

trend allows some intact fruits to be deposited away from

the immediacy, but still within the neighborhood, of the

parents. Local environmental factors also played a role

in shaping the spatial dispersal pattern (Westelaken and

Maun, 1985). Being in the downslope directions and with

the predominant wind during the summer mainly from

the south and southeast, north-bearing directions received

more intact fruits. Thus, the dispersal pattern of Taiwan

sassafras was clearly anisotropic, a fact which might

subsequently affect the next generation spatial structure.

We did not detect any size difference with respect to

distance and direction. For Taiwan sassafras, like many

bird-dispersal species, fruit size is likely to be constrained

by its long-distance dispersers (Jordano, 2000). Though

the total numbers of fruits were about the same for both

years, the proportion of mature fruits collected was higher

in 1999. This temporal variation was due to the presence

or absence of typhoons. During the dispersal season of

1999, no typhoon visited the study area. In contrast, in

early July of 2000, a severe typhoon hit the study site

directly and caused many fruits to fall before maturing

(Figure 2).

Table 3. Average size and fresh weight of mature and immature

Taiwan sassafras (Sassafras randaiense) fruits collected in

1999 and 2000.

Year

Fruit type Mean size ¡Ó 1

SE (mm)

Mean fresh

weight ¡Ó 1 SE (g)

1999

Mature

6.99¡Ó0.07 0.16¡Ó0.007

Immature 6.38¡Ó0.06 0.09¡Ó0.005

Difference 0.61¡Ó0.09

P

**

P

0.07¡Ó0.008

P

**

P

2000

Mature

6.25¡Ó0.10 0.14¡Ó0.007

Immature 5.77¡Ó0.06 0.07¡Ó0.004

Difference 0.48¡Ó0.12

P

**

P

0.07¡Ó0.007

P

**

P

Combined Mature

6.78¡Ó0.06 0.15¡Ó0.005

Immature 6.08¡Ó0.04 0.08¡Ó0.003

Difference 0.70¡Ó0.07

P

**

P

0.07¡Ó0.06

P

**

P

**

p<0.0001 based on t-test.

Figure 3. For the collected intact Taiwan sassafras (Sassafras

randaiense) fruits, observed fruit dispersal patterns in number

with respect to the distance from parents for 1999, 2000, and

both years combined.

Figure 4. For the collected intact Taiwan sassafras (Sassafras

randaiense) fruits, observed dispersal patterns in number with

respect to the eight directions for 1999, 2000, and both years

combined. F or each direction, heavy dashed-lines repres ent

the distance from the collection center, solid lines represent the

number of fruits collected, and the light dashed-lines represent

the number of fruits at a 50-fruit interval.

Table 4. Results of the likelihood ratio tests (Wald¡¦s Type 3)

of the fitted negative binomial model

a

. The dependent variable

was the number of intact Taiwan Sassafras (Sassafras ran-

daiense) fruits collected in 1999 and 2000, and the independent

variables were direction, distance, and their interaction.

Source

DF £q

P

2

P

value (p-value)

Distance

1 20.73 (< .0001)

Direction

7 42.86 (< .0001)

Distance ¡Ñ Distance

1

4.25 (0.04)

Distance ¡Ñ Direction

7 51.48 (< .0001)

a

Model overall deviance is 57.74 with df = 47 and average

deviance = 1.22; Pearson £q

2

is 52.26 with df = 47 and average

Pearson £q

2

= 1.11.

432

Botanical Studies, Vol. 47, 2006

Most of the heavier intact fruits that we collected were

deposited close to the parents, likely due to gravity. How-

ever, some were also deposited beyond the crown (Figure

6), and we considered those fruits to have been dispersed

by wind. Lin et al. (2003) found that in the study area,

probably as an adaptation to low light availability in a

cloud forest, the chlorophyll content of the outer portion

of Taiwan sassafras crown is about 22% higher than the

inner portion. Their findings suggest the outer portion of

the crown could produce heavier fruits. Though the fruits

matured, they were still light-weighted (averaged 0.15 g).

With the help of frequent strong gusts during the summer,

heavier mature fruits borne by the outer portion of the

crown could thus have the opportunity to be dispersed

away from the parents.

Our results suggest that typhoons could also affect the

dispersal of Taiwan sassafras. Long-term records (from

1897 to 1996) show that about 1.4 typhoons per year

hit the study site directly during the dispersal season.

Depending on the timing, typhoons could potentially be

both a positive and a negative factor. If a typhoon hits

the study site in mid- to late-summer, the strong winds

of typhoons could carry mature fruits away from the

maternal trees. If a typhoon occurs in early summer, as in

2000, then most of the fruits will likely to fall prematurely,

though the already matured ones could still benefit from

the visit.

Although the germination rate of intact mature Taiwan

sassafras fruits is low, such fruits might still contribute

to maintaining the local population if they could counter

the low germination rate by large numbers and could be

dispersed away from the immediacy of their parents. At

least for Taiwan sassafras, our results suggest that such a

dispersal mode is possible. Existing literature on within-

habitat dispersal of fleshy-fruited species rarely focuses

on intact mature fruits and the role of wind in dispersing

them. Our study suggests that both should deserve a closer

look.

Because we did not monitor the fate of the dispersed

mature fruits, our study could only indirectly support the

Figure 5. Observed (solid line) and fitted (dashed line) fruit

density (fruits m

-2

) with respect to the directions and the distance

from the focal trees for the intact Taiwan sassafras (Sassafras

randaiense) collected in 1999 and 2000.

Some studies have suggested that the amount of fruits

and fruit weight are positively related to maternal size

(tree height or DBH, e.g., Greene and Johnson, 1994),

and the dispersal patterns are also significantly affected

by maternal characteristics (e.g., Augspurger, 1983;

Thiede and Augspurger, 1996; Peres and Baider, 1997).

We did not detect such relationships in this study, except

in 1999, when individual tree height had a significant

and strong positive correlation with the total numbers of

fruits collected. One possible explanation is the five trees

all had an isolated and full crown, which probably means

tree fruit production was not limited by scarce resources.

The severe typhoon that occurred in early July of 2000

might also have caused the differences in fruit size and

weight between the two years (Table 3). In addition, the

spring of 1999 was the driest spring ever recorded in the

study region, with only a quarter to one-third the normal

precipitation. Whether that spring drought had any effect

on fruit weight remains to be clarified. We could not offer

any explanation why tree age had such a positive and

significant correlation with the number of damaged fruits

collected in 2000.

Figure 6. Observed (solid circle ¡Ó 1 SE) and fitted (solid line;

together with a 95% confidence interval, das hed line) mean

weight curves with respect to the distance from the focal trees

for the intact Taiwan s as safras (Sassafras randaiense) fruits

collected in 1999 and 2000.

GUAN et al. ¡X Short-distance dispersal of Taiwan sassafras

433

density-dependent hypothesis of Janzen-Connell and the

escape hypothesis of Howe and Smallwood. If fruit weight

has a positive effect on seedling emergence and survival

ability as many studies have suggested (e.g., Wenny,

2000b; Cordazzo, 2002; Gomez, 2004), then our results

agreed with the predictions made by the two hypotheses.

Yet, heavier fruits (seeds) may not guarantee better

germination success (Paz et al., 1999; Wenny, 2000b;

Gomez, 2004).

Though only accounting for a minor portion among

the fruits collected, digested fruits could be significant

in maintaining the local population. Seeds from fruits

consumed by animals usually have a higher likelihood

of germination and survival (e.g., Wenny, 2000a; but see

Wotton, 2002). The digested fruits that we collected could

be from one of the focal trees or nearby trees as suggested

by other studies (e.g., Wenny, 2000b), but they could also

be from remote seed sources. We were unable to trace

their origin.

This study was motivated by the observation that in the

study region, once established, Taiwan sassafras can ex-

pand its population, become dominant, and maintain that

dominance. We initially attributed such a phenomenon to

clonal propagation (e.g., root suckering), as in American

sassafras (Sassafras albidum) (Duncan, 1935). However,

excavations of the sapling roots showed that this was not

the case. Thus, an existing population has to rely on either

a constant input of seeds via long-distance dispersal, or via

short-distance dispersal from existing local individuals, or

both, to maintain itself and expand. Two possible modes

exist for short-distance dispersal, namely, by birds or

possibly by wind. For the second mode to work, some

quality intact mature fruits have to be deposited beyond

the immediate vicinity of the parents. Results from this

study suggest that for Taiwan sassafras the wind dispersal

mode was indeed possible. Unquestionably, constant input

of seeds from other areas can be an important source of

recruitment too. We are currently conducting a fine-scale

genetic analysis to determine the relative importance of

the two possibilities.

Acknowledgments. This study was funded by

grants from the Council of Agriculture, ROC (COA-

CF-88-10-36) and the National Science Council, ROC

(NSC89-2621-B-002-034-A10). Field supports by Mr. J.

L. Lin and the crews from the Chilan Shan Forest District

of the Forest Conservation Agency, Taiwan are deeply

appreciated. Comments from two anonymous reviewers

are also greatly appreciated.

LITERATURE CITED

Allis on, P.D. 1999. Logis tic Regression Us ing the SASP P

R

System: Theory and Application. SAS Institute Inc., Cary,

NC., pp. 217-231.

Augspurger, C.K. 1983. Seedling survival of tropical species:

interaction and dispersal distance, and pathogens. Ecology

65: 1705-1712.

Connell, J.H. 1971. On the role of natural enemies in preventing

competi tive exclus ion in some marine anim als a nd in

rain forest trees. In P.J. Den Boer and G. Gradwell (eds.),

Dynami cs of P opula tions , PUDOC, Wageni ngen, pp.

298-312.

Cordazzo, C.V. 2002. Effect of seed mass on germination and

growth in three dominant species in southern Brazilian

coastal dunes. Braz. J. Biol. 62: 427-435.

Duncan, W.H. 1935. Root systems of woody plants of old fields

of Indiana. Ecology 16: 554-567.

Godoy, J.A. and P. Jordano. 2001. Seed dispersal by animals:

exact identification of source trees with endocarp DNA

microsatellites. Mol. Ecol. 10: 2275-2283.

Gom ez, J .M. 2004. Bigger is not always better: conflicting

selective pressures on seed size in Quercus ilex. Ecology

58: 71-80.

Greene, D.F. and E.A. Johns on. 1994. Es tim ating the mean

annual seed production of trees. Ecology 75: 642-647.

Harper, J. 1977. Population Biology of Plants. Academic Press,

London, 892 pp.

Howe, H.F. and J. Smallwood. 1982. Ecology of seed dispersal.

Ann. Rev. Ecol. Syst. 13: 201-228.

Hsieh, C.F., C.F. S he n, and K.C. Yang. 1994. Introduction

to the Flora of Taiwan 3: floristic, phytogeography, and

vegetation. In Ed. Comm. Flora of Taiwan 2nd ed. Flora of

Taiwan Vol. I, 2nd ed. Taipei, Taiwan, pp. 7-16.

Hu, C.-Y. and E.-R. Ku. 1980. Observational report on fruits and

seeds of Taiwan sassafras. Taiwan For. J. 6(5): 29-31 (in

Chinese).

Janzen, D.H. 1970. Herbivores and the number of tree species in

tropical forests. Am. Nat. 104: 501-528.

Jordano, P. 2000. Fruits and F rugivory. In M. Fenner (ed.),

Seeds: The Ecology of Regeneration in Plant Communities,

2nd ed. CABI Publishing, New York, pp. 125-165.

Kollmann, J. 2000. Dispersal of fleshy-fruited species: a matter

of spatial scale. Perspect. Plant Ecol. Evol. Syst. 3: 29-51.

Levin, S.A., H.C. Muller-Landau, R. Nathan, and J. Chave.

2003. T he ecology and evolut ion of s eed dis pers al: a

theoretical perspective. Annu. Ecol. Evol. Syst. 34:

575-604.

Liao, J.C. 1997. Lauraceae. In ed. Comm. Flora of Taiwan 2nd

ed. Flora of Taiwan Vol. II, 2nd ed. Taipei, Taiwan, pp.

433-499.

Lin, J.H., B.T. Guan, S.T. Lin, and C.F. Yu. 2003. Attributes

of leave s on reproducti ve shoot s of Tai wa n sas s afras

(Sas safras randaie ns e (Hay.) Rehder) at Chilan S han,

northeastern Taiwan. J. Nat. Taiwan Univ. Exp. For. 17:

25-32 (in Chinese with English abstract).

Lin, T.P. 1992. Breaking through the bottlenecks for seedling

production of Taiwan sassafras. Taiwan For. J. 18(1): 14-16

(in Chinese with English abstract).

Neter, J., M.H. Kunter, C.J. Nachtsheim, and W. Was serman.

1996. Applied Linear S tatistical Models. 4th ed. Irwin,

434

Botanical Studies, Vol. 47, 2006

Chicago, 1408 pp.

Paz, H., S.J. Mazer, and M. Martinez-Ramos. 1999. Seed mass,

seedling emergency and environmental factors in seven rain

forest Psychotria (Rubiaceae). Ecology 80: 1594-1606.

P ere s , C .A. a nd C. Ba ide r. 1997. S ee d dis pe rsa l, s pat ial

dis tribution and population s tructure of Brazilnut trees

(Bertholletia excelsa) in southeastern Amazonia. J. Trop.

Ecol. 13: 595-616.

SAS Institute Inc., 2000. SAS/STAT

R

User¡¦s Guide, Ver. 8. SAS

Institute Inc., Cary, NC.

Stiles, E.W. 2000. Animals as seed dispersers. In M. Fenner

(ed .), S ee ds : T he E co logy of Re gene ra tion in P l ant

Communities, 2nd ed. CABI Publishing, New York, pp.

111-124.

Thiede, D.A. and C. K. Augspurger. 1996. Intraspecific variation

in seed dispersion of Lepidium campestre (Brassicaceae).

Amer. J. Bot. 83: 856-866.

Tsai, J.S. 2000. Fruit Removal Pattern on Formosan Sassafras

(Sas safras r andaiense) by Wildlife in Chilan areas . MS

Thesis, Grad. Inst. F orestry, Nat. Taiwan Univ. Taipei,

Taiwan (in Chinese with English abstract).

Wang, P.-J., C.-C. Chiou, and C.-C. Lee. 1986. Studies on seed

germination of Taiwan sassafras. Quart. J. Chinese. For.

19(1): 31-36 (in Chinese with English abstract).

Wenny, D.G. 2000a. Seed dispersal, seed predation, and seedling

recruitment of a neotropical montane tree. Ecol. Monogr.

70: 331-351.

Wenny, D.G. 2000b. Seed dispersal of a high quality fruit by

specialized frugivores: High quality dispersal. Biotropica

32: 327-337.

Westelaken, I. and M.A. Maun. 1985. Spatial pattern and seed

dis persal of Lithospermum caroliniense on Lake Huron

sand dunes. Can. J. Bot. 63: 125-132.

Wills on, M.F. and A. Travese t. 2000. The ecology of se ed

dis persa l. In M. F enner (ed.), S eeds : The E cology of

Re ge ner ati on i n P l ant Co mm uni ti es , 2 nd e d. CAB I

Publishing, New York, pp. 85-110.

Wotton, D.M. 2002. Effectiveness of the common gecko (Hoplo-

dactylus maculatus) as a seed disperser on Mana Island,

New Zealand. New Zealand J. Bot. 40: 639-647.